基于神经网络在线学习的输电线路多时间尺度负载能力动态预测

Dynamic Prediction of Multi-time Scale Load Capacity for Power Transmission Lines Based on the Online Learning of Neural Networks

输电线路负载能力的动态预测能够为负荷调度方案和线路故障检修管理提供十分重要的参考。提出基于运行环境变化对输电线路多时间尺度负载能力进行预测的方法。利用Elman神经网络结合气温、风速、负荷的历史值实现在线学习并预测气象参数和负荷,再基于稳态和暂态热容量计算模型对输电线路不同运行时间的允许负载能力进行动态预测。对比输电线路负荷与稳态、暂态负载能力,充分挖掘了输电线路的潜在容量,为制定电网科学调度和检修策略提供有力支撑。

Dynamic prediction of the load capacity of power transmission lines provides a very important reference for the load dispatch scheme and management of line troubleshooting. This article presents a method for predicting the multi-time scale load capacity of power transmission lines based on change of operational environment. Firstly,online learning and prediction of meteorological parameters and load are realized through Elman neural network as well as historical values of air temperature,wind speed and load. Then,permissible load capacities of the line at different operational time are predicted dynamically on the basis of the steady-state and transient-state heat capacity models. The load of the power transmission line is compared with its steady-state and transient-state load capacities. This approach makes a full use of the potential capacity of the transmission line and provides a strong support for the formulation of a scientific scheduling and maintenance strategy for the power grid.